Bump the PyTorch pin to 2.14 (pytorch#22501)

## Summary

PyTorch 2.14.0 is released, so move the pin from 2.13 to it. The tree is
mid-cycle at 1.5.0 and the pin normally follows the current stable
release.

The pin itself is a few version strings plus a re-sync of the vendored
c10 headers, which CI requires to match PyTorch's tree byte for byte.
Everything else here is a change 2.14 forces, each as its own commit.

## What 2.14 forces

**C++20 where ATen headers are compiled.** `intrusive_ptr.h` uses
`operator<=>` and `TensorBase.h` uses `requires`, neither guarded, so a
target that includes ATen cannot parse at C++17. This follows the
pattern already in the tree: six CMake targets set `CXX_STANDARD 20` for
exactly this reason. Added the Vulkan op tests to that list, and taught
the Buck wrapper to make the same per-target distinction. Non-ATen
targets, including everything the bare-metal presets build, still
compile at C++17.

**Two new AOTI entry points.** 2.14's generated wrapper calls
`aoti_torch_is_defined` and `aoti_torch_empty_strided_pinned`, neither
of which existed here. The first answers whether a tensor holds storage.
The second allocates ordinary host memory, because there is no pinned
allocator in this runtime and pinning only lets a copy to the device
overlap other work. It refuses a device other than CPU, as PyTorch's own
does.

**Derived shim spellings.** Some custom operators now reach the
fallback-kernel check under the name Inductor derives rather than the
name they are registered under. Both spellings are accepted for the
Metal ops and the CUDA int4 pack matmul.

**Build and CI repairs.** `setup.py bdist_wheel` is gone, so the macOS
source build uses the standard frontend. 2.14 is published for ROCm 7.2,
not 7.1. PyTorch now compiles at C++20, which makes CMake scan for
modules with scanners the images do not have. And a pip `cmake` in the
build environment made the image build silently produce a wheel with no
BLAS, so the image's own cmake is used and the result is now asserted
rather than assumed.

Two of these were already broken before this branch: the unpinned
`katex` install, and the `cmake` interaction.

## Behaviour changes

The re-sync is not cosmetic. `overflows()` answers differently for float
to integer casts, in both directions. Filling an int8 tensor with 127.5
was refused and now gives 127. A value at 2^63 cast to int64 was
accepted and wrapped, and is now refused.

The portable kernels reach this through `check_overflow_cast`, so
`full`, `full_like`, `fill`, `scalar_tensor`, `hardtanh`, `leaky_relu`,
`scatter` and `constant_pad_nd` inherit it. The header is a faithful
copy of upstream, so this is not ours to undo, but it should be visible
rather than buried in a header diff. Neither direction had a test; both
do now.

Two smaller ones come with it. Building a delegate sorts its
placeholders, and lifted tensor
constants were sorted as if they were user inputs, which left a constant
after one and made
any later constant insert impossible. They now sort with the parameters
and buffers.

And dropout was missing from the list of operators that take their
input's observer. It is an
identity once a model is not training, which is how a quantized model is
deployed, so measuring
it separately gave the operator after it a different scale. XNNPACK then
refused a reshape whose
two sides disagreed. Both spellings are listed, as several other
operators already are.

## Test plan

Each commit carries its own. Covering the whole change:

`compare_dirs.sh`, the header check CI runs, passes against a real
`release/2.14` checkout and fails without the re-sync. All eight
vendored headers are byte-identical to upstream; the one build-file edit
follows a file upstream moved.

The behaviour change was measured by compiling `overflows()` from both
branches, not read off the diff. Both new cases fail on the old header,
so neither is vacuous.

The regenerated import library was checked member by member: 45
advertised names against 42 before, none lost, and its archive metadata
is zeroed so the file is reproducible.

Not covered locally: the image build, the CUDA, ROCm, Qualcomm and Metal
jobs, the export suites, the Buck build, and anything needing Windows.
CI has all of it.

## Known outstanding

Three things this change does not repair. None of them is new here, and
each needs its own
change rather than being folded into a version bump.

No job links the checked-in Windows link stub with the Microsoft linker.
The cross build uses
a GNU linker, the same family that produced the archive, so it cannot
answer whether the
Microsoft one accepts it. That needs a Windows lowering job, which does
not exist yet. There
is now a note beside the file describing how it is produced.

The two Cortex-M model tests expect three quantize pairs where they used
to expect one. Two of
those three are round trips: the value is converted back to float and
immediately converted
again at the same scale, measured as 0.004997437 and 0.0000032501507 on
both sides. So a pass
that used to absorb them no longer matches. The pad operator those pairs
sit around is created
by this backend's own passes, after the shared pass that folds such
pairs has already run, so
absorbing them means changing the order or extending a pass another
backend shares. Correct
output, more work than needed.

Asking for pinned host memory gives ordinary host memory. There is no
pinned allocator in this
runtime, and adding one needs a matching release path, so the entry
point allocates ordinary
memory and says so. A caller that wants a copy to the device to overlap
other work will not get
the overlap.

cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils
@Sebastian-Larsson @robell @rascani

---------

Co-authored-by: PyTorch Bot <pytorchbot@users.noreply.github.com>

Author: shoumikhin

.ci/docker/ci_commit_pins/pytorch.txt

@@ -1 +1 @@
-release/2.13
+release/2.14

.ci/docker/common/install_docs_reqs.sh

@@ -20,7 +20,9 @@ if [ -n "$BUILD_DOCS" ]; then
 
   apt-get update
   apt-get install -y --no-install-recommends yarn
-  yarn global add katex --prefix /usr/local
+  # katex 0.18.5 requires commander@15 / node >= 22.12; pin to the last
+  # release compatible with the node 16 installed above
+  yarn global add katex@0.18.4 --prefix /usr/local
 
   sudo apt-get -y install doxygen
 

.ci/docker/common/install_pytorch.sh

@@ -76,21 +76,31 @@ install_pytorch_and_domains() {
   # the image compiler cannot satisfy. The venv inherits the image's
   # site-packages, so PyTorch still builds against the same numpy.
   #
-  # Keep the list in sync with pytorch/pyproject.toml [build-system].requires.
+  # Keep in sync with pytorch/pyproject.toml [build-system].requires.
   local build_venv=/tmp/pytorch-build-venv
   rm -rf "${build_venv}"
   conda_run python -m venv --system-site-packages "${build_venv}"
+  # No pip cmake: scikit-build-core would prefer it over the image's, and it
+  # searches site-packages, where MKL and libomp are not.
   conda_run "${build_venv}/bin/pip" install build "scikit-build-core>=1.0" \
-    "setuptools>=77.0.0,<82" "cmake>=3.27,<4" ninja "packaging>=24.2" \
-    "typing-extensions>=4.10.0" pyyaml six
-  conda_run "${build_venv}/bin/python" -m build --wheel --no-isolation
+    ninja "packaging>=24.2" "typing-extensions>=4.10.0" pyyaml six numpy
+  # These images have no module scanner, and nothing here uses modules.
+  conda_run env CMAKE_CXX_SCAN_FOR_MODULES=OFF \
+    "${build_venv}/bin/python" -m build --wheel --no-isolation
   rm -rf "${build_venv}"
   pip_install "$(echo dist/*.whl)"
 
+  # A build with no BLAS succeeds silently. Run from / to import the wheel.
+  (cd / && conda_run python -c "
+import torch
+assert torch._C.has_lapack, 'built without LAPACK'
+torch.linalg.qr(torch.randn(4, 4))
+")
+
   # Grab the pinned audio and vision commits from PyTorch
   TORCHAUDIO_VERSION=release/2.11
   export TORCHAUDIO_VERSION
-  TORCHVISION_VERSION=release/0.28
+  TORCHVISION_VERSION=release/0.29
   export TORCHVISION_VERSION
 
   install_domains

.ci/scripts/test-rocm-aoti.sh

@@ -7,7 +7,7 @@
 
 set -euo pipefail
 
-ROCM_VERSION="${ROCM_VERSION:-7.1}"
+ROCM_VERSION="${ROCM_VERSION:-7.2}"
 ROCM_PATH="${ROCM_PATH:-/opt/rocm}"
 PYTORCH_ROCM_INDEX="${PYTORCH_ROCM_INDEX:-https://download.pytorch.org/whl/test/rocm${ROCM_VERSION}}"
 TORCHAO_ROCM_WHEEL_BASE="${TORCHAO_ROCM_WHEEL_BASE:-https://download.pytorch.org/whl/nightly/rocm${ROCM_VERSION}}"

.ci/scripts/test-rocm-voxtral.sh

@@ -7,7 +7,7 @@
 
 set -euo pipefail
 
-ROCM_VERSION="${ROCM_VERSION:-7.1}"
+ROCM_VERSION="${ROCM_VERSION:-7.2}"
 ROCM_PATH="${ROCM_PATH:-/opt/rocm}"
 EXPECTED_ROCM_ARCH="${EXPECTED_ROCM_ARCH:-gfx950}"
 EXPECTED_WARP_SIZE="${EXPECTED_WARP_SIZE:-64}"

.ci/scripts/utils.sh

@@ -106,8 +106,8 @@ install_pytorch_and_domains() {
   local python_version=$(python -c 'import platform; v=platform.python_version_tuple(); print(f"{v[0]}{v[1]}")')
   local torch_release=$(cat version.txt)
   # Download key must match the upload key below (basename of dist/*.whl,
-  # which always carries setup.py's resolved +gitHASH). Branch-ref pins
-  # like `release/2.13` would otherwise produce `+gitrelease` here and
+  # which always carries the build's resolved +gitHASH). Branch-ref pins
+  # like `release/2.14` would otherwise produce `+gitrelease` here and
   # never hit the cache.
   local torch_short_hash=$(git rev-parse --short=7 HEAD)
   local torch_wheel_path="cached_artifacts/pytorch/executorch/pytorch_wheels/${system_name}/${python_version}"
@@ -127,18 +127,31 @@ install_pytorch_and_domains() {
   if [[ "${torch_wheel_not_found}" == "1" ]]; then
     echo "No cached wheel found, continue with building PyTorch at ${TORCH_VERSION}"
 
-    # Install PyTorch's own build-time deps so the source build does not
-    # silently inherit them from whatever else happens to be in the env
-    # (e.g. executorch's requirements-ci.txt).
-    pip install -r requirements-build.txt
     git submodule update --init --recursive
     if [[ "$(uname -m)" == "aarch64" ]]; then
       export BUILD_IGNORE_SVE_UNAVAILABLE=1
     fi
-    USE_DISTRIBUTED=1 python setup.py bdist_wheel
+    # PyTorch dropped setup.py. Build in a throwaway environment that can see the
+    # active one, so its build requirements, which pin a cmake that would be
+    # preferred over the one on PATH, cannot disturb what is installed here.
+    #
+    # Keep in sync with pytorch/pyproject.toml [build-system].requires.
+    local build_venv=/tmp/pytorch-build-venv
+    rm -rf "${build_venv}"
+    python -m venv --system-site-packages "${build_venv}"
+    "${build_venv}/bin/pip" install build "scikit-build-core>=1.0" ninja \
+      "packaging>=24.2" "typing-extensions>=4.10.0" pyyaml six numpy
+    USE_DISTRIBUTED=1 "${build_venv}/bin/python" -m build --wheel --no-isolation
+    rm -rf "${build_venv}"
     pip install "$(echo dist/*.whl)"
-
-    # Invariant: the basename setup.py just produced must match the cache
+    # A build with no BLAS succeeds silently, so check rather than assume.
+    (cd / && python -c "
+import torch
+assert torch._C.has_lapack, 'built without LAPACK'
+torch.linalg.qr(torch.randn(4, 4))
+")
+
+    # Invariant: the basename the build just produced must match the cache
     # URL we'd reconstruct on the next run. If they diverge (someone edits
     # torch_wheel_name above, or PyTorch renames its wheels), the cache
     # will silently miss and every macOS run will fall back to a ~30-min
@@ -178,7 +191,7 @@ install_pytorch_and_domains() {
   # Grab the pinned audio and vision commits from PyTorch
   TORCHAUDIO_VERSION=release/2.11
   export TORCHAUDIO_VERSION
-  TORCHVISION_VERSION=release/0.28
+  TORCHVISION_VERSION=release/0.29
   export TORCHVISION_VERSION
 
   install_domains

.ci/scripts/wheel/test_cpp_sdk.py

@@ -991,7 +991,7 @@ def test_every_shipped_header_compiles(work_dir: Path) -> None:
         # These say in their own text that they must not be included directly, and name the header to
         # include instead. Including one anyway is a use error rather than a packaging defect.
         "c10/util/complex_math.h",
-        "c10/util/complex_utils.h",
+        "torch/headeronly/util/complex_utils.h",
     )
 
     source = work_dir / "header_probe.cpp"

.github/workflows/rocm.yml

@@ -169,7 +169,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        rocm-version: ["7.1"]
+        rocm-version: ["7.2"]
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     permissions:
       id-token: write
@@ -206,7 +206,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        rocm-version: ["7.1"]
+        rocm-version: ["7.2"]
     with:
       timeout: 180
       no-sudo: true
@@ -248,7 +248,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        rocm-version: ["7.1"]
+        rocm-version: ["7.2"]
     with:
       timeout: 180
       no-sudo: true

.github/workflows/windows-msvc.yml

@@ -91,7 +91,7 @@ jobs:
 
       - name: Install build dependencies
         shell: pwsh
-        run: python -m pip install pyyaml torch==2.13.0 --extra-index-url https://download.pytorch.org/whl/test/cpu
+        run: python -m pip install pyyaml torch==2.14.0 --extra-index-url https://download.pytorch.org/whl/test/cpu
 
       - name: Build ExecuTorch
         shell: pwsh

backends/aoti/common_shims.cpp

@@ -159,6 +159,11 @@ AOTITorchError aoti_torch_get_numel(Tensor* tensor, int64_t* ret_numel) {
   return Error::Ok;
 }
 
+AOTITorchError aoti_torch_is_defined(Tensor* tensor, bool* ret_is_defined) {
+  *ret_is_defined = tensor != nullptr;
+  return Error::Ok;
+}
+
 // Device and layout utility functions
 int32_t aoti_torch_device_type_cpu() {
   // Let's say cpu is 0 for ET as well